llama3.2 on M1 Pro (16-core GPU) + 16GB unified
M1 Pro (16-core GPU) + 16GB unified
suite suite-v1
cli 0.0.5
signed8raoNQEcn3…
Embed badgesubmitted Aug 18, 2026Workload results
| Workload | Backend | Model | decode tok/s | prefill tok/s | TTFT | p50 | p95 |
|---|---|---|---|---|---|---|---|
| chat-short | ollama@0.32.14 | llama3.2Q8_0 | 103.1tok/s | 8.77tok/s | 14,247ms | 9.4ms | 10.5ms |
| chat-long | ollama@0.32.14 | llama3.2Q8_0 | 81.35tok/s | 1,246.5tok/s | 2,529ms | 11.5ms | 14.9ms |
| concurrent-decode | ollama@0.32.14 | llama3.2Q8_0 | 97.26tok/s | — | — | 9.9ms | 12.5ms |
| agent-trace | ollama@0.32.14 | llama3.2Q8_0 | 93.14tok/s | 2,975.3tok/s | 516ms | 10.5ms | 12.9ms |
Reproduce on your machine
Same workload, same model, signed at your rig. The exact command that produced this run:
$ pipx install llm-speed && llm-speed bench --model 'llama3.2' --workload 'chat-short'
Runs in about a minute. Your number lands on the leaderboard signed and linkable. How it's measured.
Embed this run
Drop the badge into a README, blog post, or signature. Each render is a backlink to the signed result.
[](https://llm-speed.com/r/r_w-p7v--eync)Related benchmarks
- More llama3.2 benchmarks — every backend and rig that has run this model.
- More M1 Pro (16-core GPU) LLM benchmarks — every model measured on this hardware.
Provenance
- Run ID
- r_w-p7v--eync
- Fingerprint hash
- 43f99b877c2d9bba
- Public key
- 8raoNQEcn33R/v+VSYxZsgizfls+6Rnx9z8JnDkiXEs=
- Received
- 2026-08-18 23:17:20